Discriminant Function Analysis (DFA) is a statistical technique used for classification and prediction. The technique can be used across multiple applications such as estimating credit risk, calculating likelihood to respond in a marketing campaign or identifying disease types in healthcare.
In these two videos we show how Discriminant Function Analysis can be used to predict the classes of a three category outcome.
Introduction to discriminant analysis – part one
In this video we introduce the procedure and show how to configure it in IBM SPSS Statistics. In this initial run, we go through each of the output tables and explain how to interpret them and assess if the statistical assumptions have been met before finally evaluating the level of accuracy in the predictions.
Part two – Additional discriminant analysis options
In Part 2, we take a deeper dive and look at optional approaches such as changing the prior probabilities, introducing bootstrapping and using Fisher discriminant coefficients to create predictive formulae.